More or Less · Topics · Episode #174
Could You Rebuild Rippling Today? Sam's Nuclear-Missile Theory of Software Jobs
Clip from the full recording, 44:37–48:53 · download mp4 · watch at this point on YouTube
Sam asks how long Rippling's $17M, two-year build would take with AI today; Parker says money moves and regulators limit speed, and Sam argues that high-stakes businesses sell trust, not software.
Sam notes Rippling started by raising what was then a huge seed round and disappearing for two years to build connectors, and asks how many days or tokens it would take today. Parker says the seed was about $17 million, which is infinitesimal now, and that it wasn’t just connectors but payroll and device management, with regulators involved.
Parker explains where AI can run wild: places where mistakes can be pushed to production and reverted. In the parts of Rippling that connect to other systems, a bug that moves money can’t always be undone, so you must be careful. Sam cites a viral thread arguing the whole commit workflow is wrong and you should push to production constantly; Parker says that works in gaming but not where businesses meet the real world.
Sam ties it to his time at Meta: the most fun jobs let you experiment because you weren’t working on nuclear missiles, and that end of the spectrum will go to pure agents. In Parker’s end you can’t screw up, so what customers buy is trust: a real balance sheet, someone you can sue, rather than the software.
Key points
- Sam: Rippling raised a big seed round and spent two years building connectors; how fast could you do that with AI today?
- Parker: the seed was about $17 million and the build included payroll and device management that needed regulators.
- Parker: AI can run wild where everything is reversible; where a bug can move money you can't always get it back.
- Sam: a viral thread says to push to production constantly; Parker says that suits gaming, not systems tied to the real world.
- Sam: fun jobs are iterative and experimental because they're not nuclear missiles, and agents will take over that extreme.
- Sam: for Rippling, what customers pay for is trust, a real balance sheet and someone to sue.
Where they landed
Quotes
“you basically had to spend seventeen million dollars or whatever it is to get off the ground and two years of your life”— Sam
“companies that can move where you can just really let AI run wild are the companies where it's everything is reversible.”— Parker
“because you're not working on nuclear missiles, right, you have the license to experiment and move super fast, which made the job really fun.”— Sam
“what you're actually paying for and doing is trust.”— Sam
Suggested tweets
Sam Lessin to Parker Conrad: in 2026, how many days would it take to rebuild the connector suite you spent $17M and two years on? Parker: money movement isn't reversible.
Tweet this →Parker Conrad: the companies where you can let AI run wild are the ones where everything is reversible.
Tweet this →Sam Lessin's nuclear-missile theory: Meta was fun because it wasn't nuclear missiles. In the businesses where you can't screw up, "what you're actually paying for is trust."
Tweet this →All topics · Full episode #174 + transcript · Subscribe on YouTube · Spotify